The present invention relates to non-invasive testing.
The development of advanced integrated circuit devices and architectures has been spurred by the ever increasing need for speed. For example, microwave, fiber optical digital data transmission, high-speed data acquisition, and the constant push for faster digital logic in high speed computers and signal processors has created new demands on high-speed electronic instrumentation for testing purposes.
Conventional test instruments primarily include two features, the integrated circuit probe that connects the test instrument to the circuit and the test instrument itself. The integrated circuit probe has its own intrinsic bandwidth that may impose limits on the bandwidth achievable. In addition, the probe also determines an instrument's ability to probe the integrated circuit due to its size (limiting its spatial resolution) and influence on circuit performance (loading of the circuit from its characteristic and parasitic impedances.). The test instrument sets the available bandwidth given perfect integrated circuit probes or packaged circuits, and defines the type of electric test, such as measuring time or frequency response.
Connection to a test instrument begins with the external connectors, such as the 50 ohm coaxial Kelvin cable connectors (or APC-2.4). The integrated circuit probes provide the transitions from the coaxial cable to some type of contact point with a size comparable to an integrated circuit bond pad. Low-frequency signals are often connected with needle probes. At frequencies greater than several hundred megahertz these probes having increasing parasitic impedances, principally due to shunt capacitance from fringing fields and series inductance from long, thin needles. The parasitic impedances and the relatively large probe size compared to integrated circuit interconnects limit their effective use to low-frequency external input or output circuit responses at the bond pads.
Therefore, electrical probes suffer from a measurement dilemma. Good high-frequency probes use transmission lines to control the line impedance from the coaxial transition to the integrated circuit bond pad to reduce parasitic impedances. The low characteristic impedance of such lines limits their use to input/output connections. High-impedance probes suitable for probing intermediate circuit nodes have significant parasitic impedances at microwave frequencies, severely perturbing the circuit operation and affecting the measurement accuracy. In both cases, the probe size is large compared to integrated circuit interconnect size, limiting their use to test points the size of bond pads. Likewise sampling oscilloscopes, spectrum analyzers, and network analyzers rely on connectors and integrated circuit probes, limiting their ability to probe an integrated circuit to its external response. For network analysis, a further issue is de-embedding the device parameters from the connector and circuit fixture response, a task which grows progressively more difficult at increasing frequencies.
With the objective of either increased bandwidth or internal integrated circuit testing with high spatial resolution (or both) different techniques have been introduced. Scanning electron microscopes or E-beam probing uses an electron beam to stimulate secondary electron emission from surface metallization. The detected signal is small for integrated circuits voltage levels. The system's time resolution is set by gating the E-beam from the thermionic cathodes of standard SEM's. For decreasing the electron beam duration required for increased time resolution, the average beam current decreases, degrading measurement sensitivity and limiting practical systems to a time resolution of several hundred picoseconds. Also, SEM testing is complex and relatively expensive.
Valdmanis et al., in a paper entitled “Picosecond Electronics and Optoelectronics”, New York: Springer-Verlag, 1987, shows an electro-optic sampling technique which uses an electrooptic light modulator to intensity modulate a probe beam in proportion to a circuit voltage. Referring to FIG. 1, an integrated circuit 10 includes bonded electrical conductors 12 fabricated thereon whereby imposing differential voltages thereon gives rise to an electric field 14. For carrying out a measurement an electro-opti needle probe 16 includes an electro-optic tip 18 (LiTaO3) and a fused silica support 20. A light beam incident along path 22 is reflected at the end of the electro-optic tip 18 and then passes back along path 24. An electric field 14 alters the refractive index of the electro-optic tip 18 and thereby alters the polarization of the reflected light beam on the exit path 24, which thus provides a measure of the voltages on the conductors 12 at a single point. Unfortunately, because of the proximity of the probe 16 to the substrate 10 capacitive loading is applied to the circuit, thereby altering measurements therefrom. In addition, it is difficult to position the probe 16 in relation to the conductor because the probe 16 and circuit 10 are vibration sensitive. Also, the measurements are limited to conductors 12 on or near the surface of the circuit 10. Further, the circuit must be active to obtain meaningful results and the system infers what is occurring in other portions of the circuit by a local measurement.
Weingarten et al. in a paper entitled, “Picosecond Optical Sampling of GaAs Integrated Circuits”, IEEE Journal of Quantum Electronics, Vol. 24, No. 2, February 1988, disclosed an electro-optic sampling technique that measures voltages arising from within the substrate. Referring to FIG. 2, the system 30 includes a mode-locked Nd:YAG laser 32 that provides picosecond-range light pulses after passage through a pulse compressor 34. The compressed pulses are passed through a polarizing beam splitter 36, and first and second wave plates 38 and 40 to establish polarization. The polarized light is then directed at normal incidence onto an integrated circuit substrate 42. The pulsed compressed beam can be focused either onto the probed conductor itself (backside probing) or onto the ground plane beneath and adjacent to the probed conductor (front-side probing). The reflected light from the substrate is diverted by the polarizing beam splitter 36 and detected by a single point slow photo diode detector 44. The photo diode detector is also connected to a display 46.
A microwave generator 48 drives the substrate 42 and is also connected to an RF synthesizer 50, which in turn is connected to a timing stabilizer 52. The pulse output of the laser 32 is likewise connected to the timing stabilizer 52. The output of the stabilizer 52 connects back to the laser 32 so that the frequency of the microwave generator 48 locks onto a frequency that is a multiple of the laser repetition rate plus an offset. As a consequence, one may analyze the electric fields produced within the integrated circuit as a result of being voltage drive, thus providing circuit analysis of the integrated circuit operation. In essence, the voltage of the substrate imposed by the microwave generator 48 will change the polarization in the return signal which results in a detectable change at the diode detector 44.
Referring to FIGS. 3A and 3B, the locations along the incident beam are designated a, b, c (relative to the “down” arrow), and designated along the reflected beam as d, e, and f (relative to the “up” arrow), and the intensity modulated output signal is designated as g. The corresponding states of polarization exhibited in the measurement process are shown in the similarly lettered graphs of FIG. 3B. At location a of FIG. 3A, the polarizing beam splitter 36 provides a linearly polarized probe beam (as shown in graph a of FIG. 3B) that is passed through the first wave plate 38, which is a T/2 plate oriented at 22.5 degrees relative to the incident beam polarization, so as to yield at location b the 22.5 degree elliptically polarized beam shown in graph b of FIG. 3B). The beam then passes through the second wave plate 40, which is a T/2 plate oriented at 33.75 degrees relative to the incident beam, so as to rotate the beam an additional 22.5 degrees to yield at location c the 45 degree polarization (shown in graph c of FIG. 3B), which is at 45 degrees to the [011] direction of the substrate 42, i.e., the cleave plane of the wafer. Similar rotations are shown for the reflected beam at the successive locations d, e, and f, the resultant polarizations respectively being as shown in graphs d, e, and f of FIG. 3B. As shown in graph f in particular, the electro-optic effect of any voltage present on the substrate 42 at the spot at which the beam reflects therefrom brings about a change in the specific polarization orientation in an amount designated in graph f of FIG. 3B as &, and that change is reflected in an amplitude change or intensity modulation in the output signal at location g that passes to the photo-diode 44 (as shown in graph g of FIG. 3B). It is the measurement of & that constitutes the voltage measurement. Among the various techniques of pre-determining the voltage patterns to be used in testing an integrated circuit, or indeed an entire printed circuit, Springer, U.S. Pat. No. 4,625,313, describes the use in a CPU of a ROM “kernel” in which are stored both a test program sequence and the testing data itself.
Since the system taught by Weingarten et al. does not include a probe proximate the circuit under test the limitations imposed by capacitive loading of the circuit to be tested is avoided. However, the system taught by Weingarten et al. is limited to “point probing,” by the lens 41 converging the input beam into a test point on the order of one wavelength. Unfortunately, to test an entire circuit an excessive number of tests must be performed. In addition, it is not possible to test multiple points simultaneously without the use of multiple systems, which may be useful in testing different portions of the circuit that are dependant upon one another. The resulting data from the system is presented to the user as a single amplitude measurement, i.e., the intensity of the signal produced at the photo-diode 44 depends simply upon the degree to which the polarization of the reflected light entering the beam splitter 36 has been rotated, so that not only are the actual phase and polarization data that derive the reflection process lost, but the precision and accuracy of the measurement becomes subject to the linearity and other properties of the photo-diode 44 and the display 46.
Various other techniques by which semiconductors may be characterized, using electromagnetic radiation of different wavelengths under different conditions is cataloged by Palik et al. in “Nondestructive Evaluation of Semiconductor Materials and Device,” Plenum Press, New York, 1979, chapter 7, pp. 328-390. Specifically, treatment is given of (1) infrared reflection of GaAs to obtain the optical parameters n and k and then the carrier density N and mobility u; (2) infrared transmission in GaAs to determine k from which is determined the wavelength dependence of free carrier absorption; (3) infrared reflection laser (spot size) scanning of and transmission through GaAs to determine free carrier density in homogeneity, including local mode vibrations; (4) far infrared impurity spectra; (5) infrared reflection and transmission from thin films on a GaAs substrate; microwave magnetoplasma reflection and transmission; (6) submillimeter-wave cyclotron resonance in GaAs to determine magnetotransmission; (7) ruby laser radiation to form a waveguide in a GaAs film on a GaAs substrate, the propagation features of which are then measured using infrared radiation; (8) infrared reflectance from multilayers of GaAs on a GaAs substrate; (9) reflectance measurements of graded free carrier plasmas in both PbSnTe films on PbSnTe substrates and InAs on GaAs substrates; (10) interferometric measurements of ion implanted layers; (11) infrared restrahlen spectra, also to determine lattice damage effects; (13) ellipsometric measurements of ion-implanted GaP; (14) determination of optical constants by internal reflection spectroscopy; (15) laser raster scanning of semiconductor devices to measure photoconductivity, to track the flow of logic in a MOS shift register (because of current saturation, the effect of the laser light differs in cells in the 0 or 1 logic state), and with a more intense laser power level to change those logic states (i.e., to write to the circuit); (16) laser raster scanning of semiconductor devices to determine variations in resistivity and carrier lifetimes; (17) thermal imaging of circuits to find hot spots; (18) Raman backscattering to determine free carrier density; (19) carrier injection to study the band edge; (20) birefringence measurements in monolayers of GaAs and AlAs on GaAs to characterize the resultant strain; (21) photoluminescence and cathodoluminescence measurements of implanted layers and acceptor and donor densities. With the exception of (7) above which relates to waveguide transmission, and also of (15) and (17), these techniques relate to the characterization of static systems. While (15) relates to a spot canning technique of the operational integrated circuit and (17) relates to hot-characterization of the device temperature.
What is desired, therefore, is a high bandwidth non-invasive testing system for semi-conductor materials.